!pip3 install yfinance mplfinance mpl_finance
In [1]:
from financial_market import *

1.1 Get S&P 500 index from January 2000 to June 2022 for stocks that were present the whole time and create a table of their closing prices.

In [2]:
df_concat, df_sector_dict, symbols = load_dataframe()
df_concat
Out[2]:
symbol security sec_filings gics_sector gics_sub-industry headquarters_location date_first_added cik founded
253 IPG The Interpublic Group of Companies reports Communication Services Advertising New York City, New York 1992-10-01 51644 1961 (1930)
51 T AT&T reports Communication Services Integrated Telecommunication Services Dallas, Texas 1983-11-30 732717 1983 (1885)
470 VZ Verizon reports Communication Services Integrated Telecommunication Services New York City, New York 1983-11-30 732712 1983 (1877)
353 OMC Omnicom Group reports Communication Services Advertising New York City, New York NaN 29989 1986
361 PARA Paramount Global reports Communication Services Movies & Entertainment New York City, New York NaN 813828 2019 (1952)
... ... ... ... ... ... ... ... ... ...
117 CMS CMS Energy reports Utilities Multi-Utilities Jackson, Michigan 1999-05-03 811156 1886
11 AES AES Corporation reports Utilities Independent Power Producers & Energy Traders Arlington, Virginia 1998-10-02 874761 1981
98 CNP CenterPoint Energy reports Utilities Multi-Utilities Houston, Texas 1985-07-31 1130310 1882
378 PPL PPL Corporation reports Utilities Electric Utilities Allentown, Pennsylvania NaN 922224 1920
167 EIX Edison International reports Utilities Electric Utilities Rosemead, California 1957-03-04 827052 1886

205 rows × 9 columns

In [3]:
df, df_sector, df_both, not_done_for = get_closing_prices(symbols, df_sector_dict)

TPR not in past
- BF.B: No data found for this date range, symbol may be delisted
BF.B delisted
MET not in past
HWM not in past
WRK not in past
PARA not in past
In [4]:
# these have become nan because symbols has values that aren't collected
print(df.isnull().values.any(), not_done_for)

False ['BF.B']
In [5]:
print(df.columns[df.isna().any()].tolist())

[]
In [6]:
df.to_excel("output/excel/sp500_stock_names.xlsx")
df_sector.to_excel("output/excel/sp500_sector_names.xlsx")
df_both.to_excel("output/excel/sp500_both.xlsx")
In [7]:
df.to_csv("output/csv/sp500_stock_names.csv")
df_sector.to_csv("output/csv/sp500_sector_names.csv")
df_both.to_csv("output/csv/sp500_both.csv")
In [8]:
# here you can change the stock to be plotted to any other
# ticker value that you want to be plotted
stock_to_be_plotted = "AAPL"
plot_candlesticks(stock_to_be_plotted)

[*********************100%***********************]  1 of 1 completed
In [9]:
plt.plot(df["date"], df["AAPL"])
Out[9]:
[<matplotlib.lines.Line2D at 0x7fef05767640>]

1.2 Plot all the closing price time series in nx3 subplots.

Here you can clearly see the stock name as well as the sector the stock belongs to (format: stock_sector)

In [10]:
df_both = pd.read_csv("output/csv/sp500_both.csv", index_col=None)
df_both = df_both.iloc[:, 1:]
df_both["date"] = df_both['date'].apply(pd.to_datetime)
In [11]:
plot_all_charts(df_both)

1.3 Construct a correlation matrix for full-time using log returns r(t) = ln(P(t)) - ln(P(t-1)) and plot it.

In [12]:
# df = pd.read_excel("output/excel/sp500_sector_names.xlsx", sheet_name="Sheet1", index_col=None)
df = pd.read_csv("output/csv/sp500_sector_names.csv", index_col=None)
df = df.iloc[:, 2:]
# this is without calculating the log returns.
plot_corr_mat(df)
In [13]:
# calculating log returns and then plotting
for i in df.columns:
    df[i] = np.log(df[i]/df[i].shift(1))
df = df.iloc[1:, :]
plot_corr_mat(df)

1.4 Construct correlation matrices using r(t) of epoch size 400 working days and with a shift of 200 days.

In [14]:
dict_col = ticker_val_locations(df)
fig, axes = plt.subplots(len(df) // (200 * 3), 3, figsize=(20, 5 * (len(df) // (200 * 3))))
count = 0
for i in range(0, len(df) - 200, 200):
    try:
        p = sns.heatmap(df.iloc[i:i+400,:].corr(), ax=axes[count//3, count%3])
        p.set_xticks(ticks=list(dict_col.values()), labels=list(dict_col.keys()))
        p.set_yticks(ticks=list(dict_col.values()), labels=list(dict_col.keys()))
        count += 1
    except:
        pass